A Genome-Wide Association Study of Equine Metabolic Syndrome And

A GENOME-WIDE ASSOCIATION STUDY OF EQUINE METABOLIC SYNDROME AND

PITUITARY PARS INTERMEDIA DYSFUNCTION

A Thesis

Presented to the Faculty of the Graduate School of Cornell University In Partial Fulfillment of the Requirements for the Degree of Master of Science

by Cassandra Lois Streeter January 2012

ABSTRACT Laminitis remains a poorly understood form of lameness in the horse despite being heavily researched. Endocrinopathic laminitis is an insidious form of laminitis that often results from insulin resistance (an indicator of Equine Metabolic Syndrome or EMS) or equine pituitary pars intermedia dysfunction (PPID). Both conditions occur in older horses with PPID having a later age of onset than EMS and horses are often characterized as having a “cresty” neck. EMS horses have persistent hyperinsulinemia while horses affected by PPID have elevated ACTH levels. Though no specific genetic predispositions have been identified for EMS or PPID, ponies are frequently observed to have both conditions. The objective of this study was to conduct a genome-wide association to identify candidate genes that may predispose individuals to EMS, PPID or both. A total of 65 horses, at least ten years old and of full or majority Arabian descent were divided into four categories based on previous endocrinology testing. Horses with ACTH levels greater than 70 pg/mL and insulin levels less than 40 uIU/mL were designated “PPID” while horses with insulin levels > 70 uIU/mL and ACTH levels less than 40 pg/mL were designated “EMS”. Horses with ACTH and insulin levels < 40 pg/mL or uIU/mL, respectively, were considered “Normal” and those with both levels > 70 were grouped into the fourth category termed “Both”. The Illumina Infinium® II Assay (Illumina Inc, San Diego, CA, USA) was performed on the EquineSNP 50 Genotyping BeadChip (Illumina Inc, San Diego, CA, USA) using DNA extracted from tail hair or blood samples. Data was analyzed using the PLINK v1.07

Whole genome association analysis toolset and JMP 8.0 (SAS Insitute Inc, Cary, NC, USA). Basic association (chi-square) tests were used to compare various groupings of disease cases and controls for a total of 15 qualitative associations. Quantitative associations were also performed for highest recorded ACTH (excluding August-October measurements) and insulin levels. An additional association was performed comparing horses with at least one episode of laminitis versus those with no recorded history of laminitis, regardless of disease category. SNPs with a missing genotype rate greater than 10% were excluded from analysis. SNPs in linkage disequilibrium (r2 > 0.99) were pruned to a single SNP per haplotype block in all graphs. Any SNPs exceeding the Bonferroni α threshold were mapped on the UCSC Genome Browser using the September 2007 assembly (EquCab2.0) of the horse genome. Candidate genes were identified within ~50 kb of each significant SNP. A total of 68 SNPs representing 59 different loci exceeded the significance threshold after correction for multiple testing. The most significant SNP, BIEC2-215377 (ECA 13) with a P-value of 4.34 e-7, was identified in the quantitative association with the highest ACTH level recorded for each individual. This SNP yielded one candidate gene, XPO6. Other SNPs that did not exceed the significance threshold but did vary from their expected P-value in a quantile-quantile plot include BIEC2-770354 (ECA3) and its neighboring SNP in linkage disequilibrium, BIEC2-770355 which yielded two candidate genes, FTO and ATP5H. This study resulted in the discovery of several good candidate genes that are worthy of fine mapping. However, epistatic effects may hamper the ability to identify all genetic predispositions. Improvements in diagnostic testing may allow for more specific classification of disease categories and refinement of the current data set. The present study is also limited by the relatively limited coverage of the Equine SNP50 chip. Expanded platforms with better coverage may allow for the discovery of more candidate genes.

BIOGRAPHICAL SKETCH Cassy Streeter was born and raised in Syracuse, NY. She received her B.S. in Animal Science at Cornell University in January 2006. She is a Thoroughbred and Standardbred racing person.

After this thesis she now considers herself an Arabian person and, after receiving many offers for free Arabians during the study, she may end up owning one in the future.

iii

This thesis is dedicated to all of the owners of the horses that participated in this study.

iv ACKNOWLEDGMENTS I would like to thank my advisor, Dr. Samantha Brooks, as well as my other committee members, Dr. Yves Boisclair and Dr. Ned Place, for taking the time to review my thesis. I would also like to thank Beth Minnich and the Arabian Horse Foundation as well as Dr. Don Walsh and the Animal Health Foundation for their support of this project. Thank you to Barb Schanbacher and the Endocrinology lab for their help collecting DNA samples and health records. I extend my undying gratitude to Bethany for waking up at the crack of dawn in the middle of winter to help me draw blood from shaggy Arabians. My undergraduates, Elissa,

Lauren, Carla, Brooke, Nicole, Rachel and Taylor, deserve a big thank you for testing my patience, supplying funny stories and making me a better person. It was a real pleasure passing on my knowledge and learning new things along with them. They remind me how fun science can really be.

I’d like to express my sincere appreciation to all of the owners and horses who participated in this study. Their personal stories were a constant reminder of why this work like this needs to be done. I would also like to thank Jim Moran and Angelo LaBella for allowing me to be their “barn mascot” and feel like I actually own horses even though I don’t foot the bills.

Finally, I’d like to thank my parents, sister and brother for all of their love and support throughout the years.

v TABLE OF CONTENTS

Biographical sketch iii Dedication iv Acknowledgements v List of Figures vii List of Abbreviations viii Chapter 1 – Review of Literature 1 Chapter 2 – Original Research 21 Appendix 46

vi LIST OF FIGURES Figure 2.1 Quantitative association of highest recorded ACTH value p 27 Figure 2.2 Quantitative association of highest recorded insulin value p 28

Figure 2.3 Qualitative association of hyperinsulinemia (B or M) vs. controls (N) p 29 Figure 2.4 Quantile-quantile plot of qualitative association of all disease groups (B, M and P) vs. controls (N) p 30

vii LIST OF ABBREVIATIONS α-MSH – alpha-melanocyte-stimulating hormone ACTH – adrenocorticotropic hormone

AIRg – acute insulin response to glucose BCS – body condition score DST – dexamethasone suppression test ECD – equine Cushing’s Disease EMS – equine metabolic syndrome

FSIGTT – frequently sampled intravenous glucose tolerance test POMC – proopiomelanocortin PPID – pituitary pars intermedia dysfunction SNP – single nucleotide polymorphism

viii CHAPTER 1 Review of Literature Introduction

Laminitis remains a poorly understood form of lameness in the horse despite being heavily researched. While many studies have focused on the pathophysiology of acute onset laminitis due to experimental models like carbohydrate overload [1] and black walnut extract [2], attention is now being given to an insidious form of laminitis referred to as endocrinopathic laminitis [3]. Endocrinopathic laminitis is often a result of insulin resistance (associated with equine metabolic syndrome) or equine pituitary pars intermedia dysfunction (aka equine Cushing’s Disease). Both equine metabolic syndrome and pituitary pars intermedia dysfunction show similar phenotypes and can be difficult to distinguish without diagnostic testing. There are currently no cures for either condition, though some management and drug therapies are available. Equine Metabolic Syndrome The term equine metabolic syndrome (EMS) was first coined in 2002 by Johnson to describe horses that show signs of chronic/previous laminitis in the absence of acute triggers such as carbohydrate overload or colic [4]. These horses are often obese and referred to as “easy keepers” because of their ability to maintain body weight with decreased caloric intake. The core characteristics of a horse affected by

EMS are increased adiposity (often regional), insulin resistance (characterized by hyperinsulinemia) and a predisposition towards laminitis [4]. Johnson noted the age of onset to range from 8-18 years [4], though cases have occasionally appeared outside of this age range. Obesity in horses is not as clearly defined as the body mass index designed for humans [5] because the popular Henneke scale [6] does not account for regional adiposity. While generalized obesity is often linked to insulin resistance in horses [7],

1 regional adiposity is now gaining recognition as a high risk factor for insulin resistance [8, 9]. The regional adiposity often seen in an EMS affected horse manifests as a “cresty neck” in which fat accumulates along the nuchal ligament on the dorsal side of the neck. Larger mean neck circumference has previously been shown to be associated with lowered glucose tolerance [8].This regional adiposity is comparable to the accumulation of fat around the waist in humans, a characteristic that is a risk factor for human metabolic syndrome [10]. Since the Henneke scale does not properly address this issue, an objective scoring system for quantifying the relative size of the crest of the neck has been developed with scores ranging from 0 to 5 [11]. Horses with scores greater than or equal to 3 are classified as having an excessively cresty neck. Horses in this category also tend to have regional fat accumulation in the supraorbital fat pads, tail head and the shoulder area [4].

Hyperinsulinemia, an indication of insulin resistance [8, 12, 13], is a key characteristic of EMS [4]. While the plasma insulin concentration in fasted, healthy horses is often less than 20 uIU/mL [14], insulin resistant horses can have levels in excess of 70 uIU/mL [15]. Ponies with a history of laminitis were observed to have higher levels of insulin when compared to non-laminitic ponies [16]. In addition, induction of hyperinsulinemia, while maintaining euglycemia, has been shown to induce laminitis in otherwise healthy ponies and horses [17, 18]. Though both of these studies used insulin concentrations in excess of the normal physiological range (>1000 uIU/mL), the onset of laminitis in the presence of an acute increase in insulin indicates that laminitis may also be brought on by prolonged or chronic hyperinsulinemia at physiological levels [17, 18]. It’s possible that hyperinsulinemia alone does not bring about a laminitic episode; rather it predisposes insulin resistant horses to developing laminitis due to other causes. For example, acute ingestion of a large amount of

2 nonstructural carbohydrates in feed or pasture that leads to disruption of the microbial ecology of the intestine. Hyperglycemia is often seen in humans as a result of insulin resistance and β cell failure [19]. However, hyperglycemia is rarely observed in horses with insulin resistance [16, 20], though isolated cases have been reported [21]. Place et al. (2010) observed significantly higher blood glucose levels in EMS affected horses than in controls, but the levels were still within the reference range [22]. It has been proposed that resistance to hyperglycemia in horses may be due to more robust β cell function than that of humans [20]. It is also possible that horses simply do not have a long enough lifespan for β cell failure (and subsequent hyperglycemia) to develop [20]. The repeated postprandial hyperinsulinemia that results from consumption of diets high in nonstructural carbohydrates may bring about the development of insulin resistance in the horse [23] and, subsequently, hyperinsulinemia. Alternatively, hyperinsulinemia may also be due to hepatic insulin resistance where insulin clearance from the liver is lower [24]. Insulin normally acts to stimulate glucose uptake in skeletal and adipose tissue by the recruitment of the GLUT4 glucose transport protein to the cell surface. In ponies, GLUT4 stimulation is lower than that observed in pigs and cows which may account for their susceptibility to insulin resistance [25]. Insulin resistant horses in general show decreased basal levels of active cell-surface GLUT4, though overall GLUT4 content remains the same [26]. Waller et al. used a biotinylated bis-mannose photolabeled technique that was able to distinguish active cell-membrane GLUT4 [26] while other studies simply used fractionation which only determines GLUT4 content. It may be worthwhile to repeat the work done by Duehlmeier et al. to see if pony breeds have decreased active GLUT4 when compared to horses.

Equine Metabolic Syndrome Diagnostic Testing

3 Many different diagnostic tests have been investigated to assess the degree of insulin sensitivity and insulin resistance in horses with metabolic syndrome. The resting plasma concentration of insulin has been shown to be a useful diagnostic test for insulin resistance [8]. Insulin and glucose responses to feeding are higher in the morning [27], so fasting insulin levels are preferred. Insulin concentrations do not appear to be influenced seasonally [22, 28]. The cutoff for hyperinsulinemia is variable among testing labs and researchers. Values ranging from 32 uIU/mL [29] to 70 uIU/mL

[15] have been used to identify insulin resistance and EMS.

Eiler et al. developed the combined glucose insulin tolerance test in response to the need of practitioners to have a test that is easily applied to a clinical setting [30].

Horses are first fasted and baseline blood samples are drawn. Then, via catheter, dextrose followed by insulin infusions are administered and a second blood sample is drawn 45 minutes later. The blood samples are measured for both glucose and insulin levels and failure to return to baseline glucose levels by 45 minutes or the maintenance of elevated insulin levels indicates decreased insulin sensitivity [31]. The advantage of this test is that it takes less than one hour to collect all necessary samples and minimal endocrinology testing is required. However, this test may not be accurate when performed during a laminitic episode [31] and placement of catheters the night before the test is recommended.

A technique often used to assess insulin sensitivity in a research setting is the frequently sampled intravenous glucose tolerance test (FSIGTT) [32][13, 21, 26, 33, 34]. A minimal model analysis is used to determine the value of the acute insulin response to glucose (AIRg) and insulin sensitivity. Higher AIRg values have been linked with

4 lower insulin sensitivity in horses and ponies [35]. The FSIGTT involves injecting 100 mg/kg bodyweight dextrose followed by 20 mu/kg bodyweight insulin after 20 minutes [36]. Twenty-eight blood samples are then drawn at various times up to 180 minutes after the dextrose injection. Samples are measured for insulin and glucose plasma concentrations and the area under the curve for both insulin and glucose are calculated. The biggest drawback to this test is the number of samples needed and the length of time required to complete the test. However, this test is more specific to insulin sensitivity than the basal plasma insulin concentration, which only identifies the presence or absence of hyperinsulinemia. Due to its labor intensive nature, this diagnostic test is not often used in a clinical setting.

Equine Metabolic Syndrome Treatments and Therapies

Metformin is an oral biguanide used to improve insulin sensitivity and is the most widely prescribed drug for the treatment of type II diabetes in humans [37]. The efficacy of metformin in the horse is still being assessed. Vick et al. found that metformin was not effective long term for obese mares [38]. However, the higher dosage administered by Durham et al. showed improvement in insulin sensitivity up to

220 days but insulin resistance proxies failed to continue to decrease after two weeks

[20]. As recurrent episodes of laminitis are linked to decreased insulin sensitivity [16],

Durham et al. noted that metformin may be more effective prior to the first bout of laminitis [20]. The low bioavailability of metformin in the horse [39] may account for the lack of efficacy seen in previous studies and higher dosages may be needed to fully evaluate metformin as a treatment for EMS. Owners may also use chromium supplements for insulin resistant horses as it has been shown to increase mRNA

5 expression of insulin and GLUT4 receptors in the skeletal muscle of mice [40].

Pioglitazone, an insulin sensitizing drug used in humans, was recently investigated for use in horses, but no significant improvement in insulin sensitivity was detected [41].

Due to a lack of proven pharmaceutical therapies available for metabolic syndrome the most effective treatment for affected horses is intense management. As with humans, the best treatment to improve insulin sensitivity is through dietary restriction and increased exercise [42]. Limiting intake of feeds high in nonstructural carbohydrates will avoid sharp increases in plasma insulin concentrations and, therefore, the development or aggravation of insulin resistance [43]. This will also help shorten the insulin refractory time during which hyperglycemia is present [4]. Moderate exercise alone without concurrent adjustment to diet was shown to be ineffective in reducing insulin levels in obese individuals [33]. However, the combination of diet control and exercise was successful in lowering insulin levels and bodyweight [15], though results varied by individual.

Hypothyroidism and Thyroid Function in EMS horses

Horses affected by EMS are often incorrectly diagnosed as having hypothyroidism as these animals tend to have lower circulating thyroid hormone levels

[44]. However, thyroidectomized horses do not show the clinical signs of regional adiposity and laminitis seen in EMS affected individuals [45, 46]. Levothyroxine is often prescribed for EMS affected horses to improve insulin sensitivity as pretreatment of horses with levothyroxine has been shown to prevent the development of endotoxemia-induced insulin resistance [47]. Levothyroxine may also be administered

6 to raise the basal metabolic rate and help “jump start” weight loss in horses that are currently affected by laminitis where an exercise regimen cannot be implemented.

Equine Pituitary Pars Intermedia Dysfunction (Equine Cushing’s Disease)

Equine Pituitary Pars Intermedia Dysfunction (PPID), also commonly referred to as Equine Cushing’s Disease [48], is the most common endocrinopathy in late middle-aged and geriatric horses [49]. PPID results in increased production of proopiomelanocortin (POMC) derived peptides due to the loss of dopaminergic inhibition from the hypothalamus [50-53]. POMC is further processed into adrenocorticotropic hormone (ACTH), corticotrophin-like intermediate lobe peptide

(CLIP), melanocyte-stimulating hormone (MSH), β-endorphin and β-lipotropin [49].

Most PPID affected horses develop a pituitary adenoma or show signs of adenomatous hyperplasia [54, 55]. Originally, it was believed that the adenoma itself was rather inactive and it was the encroachment of the tumor on the hypothalamus that accounted for the clinical signs of Cushing’s Disease [56]. However, it has been shown that this hyperplasia is responsible for increased circulating levels of ACTH [57] and β- endorphins [58]. These horses may also develop adrenal hyperplasia [55] and, in turn, hyperadrenocorticism. Though rare, some cases of pituitary independent Cushing’s syndrome have been reported in the horse [59].

The pars intermedia, consisting of melanotropes, is under the direct inhibitory control of dopamine secreted by the periventricular neurons of the hypothalamus [60].

McFarlane et al., using tyrosine hydroxylase staining, found that the expression of dopamine in PPID horses was only 20% of the levels expressed in younger and age- matched controls [53]. These results are in agreement with a previous study that found

7 dopamine levels to be decreased 88% in PPID affected horses [58]. Knockout mice deficient in dopamine D2 receptors (expressed in the pars intermedia) show selective proliferation of the melanotropes of the intermediate lobe which then double their

POMC mRNA expression [61]. There is also alternative processing of POMC within melanotropes with increased expression of prohormone convertase 1 which, in turn, increases levels of ACTH [61]. Further, it was found that corticotropin-releasing factor, the releasing factor that stimulates corticotropes and melanotropes via the paraventricular nucleus [62], is not responsible for the increased POMC synthesis [61].

PPID shares many clinical signs with EMS including regional adiposity and a predisposition to laminitis [63]. The most obvious clinical sign of a PPID affected individual is abnormal hair coat which can manifest as hirsutism, delayed coat shedding or incomplete shedding [64, 65].Though the exact mechanism by which horses develop hirsutism is not understood, this sign has been shown to have a positive predictive value of 90% for PPID [66]. Mares may also display an abnormal estrus cycle [63]. PPID has a mean age of onset of 20 years [65, 67], though cases have been noted as young as 7 years [54]. A correlation has been shown between age and plasma cortisol and ACTH concentrations in healthy horses which suggests that age may be a predisposing factor for PPID [68]. Other signs of PPID include polyuria, polydipsia and muscle wasting [49].

Seasonal variation in plasma ACTH concentration have been shown with higher levels in the autumn [22, 28, 68], though the timing and duration of this elevation does not differ between PPID affected horses and healthy horses [69]. This seasonal increase is thought to help prepare the horse metabolically for lean winter months [70].

8 It has been reported that ponies have a lower plasma ACTH concentration than horses

[51], but Donaldson et al. showed that there is no significant difference in ACTH levels between healthy horses and ponies [68].

Equine Pituitary Pars Intermedia Dysfunction Diagnostic Testing

A popular method for detecting PPID in horses is the dexamethasone suppression test (DST). This test requires plasma cortisol concentrations to be measured from blood samples drawn before and 19 hours after intramuscular administration of dexamethasone [48]. Horses affected by PPID will fail to suppress cortisol levels below 1 ug/dL [66]. Though it was first reported that the DST has 100% sensitivity and specificity [48], it has since been shown that the specificity may be closer to 76% and sensitivity to 65% [66]. The test is less accurate during the seasonal

ACTH increase in the autumn, and the pituitary axis in older horses is less sensitive to dexamethasone administration [68]. There is also concern that the administration of dexamethasone can exacerbate laminitis in horses with current or previous bouts of laminitis [48, 49], although no laminitis was observed in a study using the DST [48].

Another complication of the dexamethasone suppression test is the need for multiple blood draws and the additional cost incurred by the owner. A cheaper option is needed, especially if there is a need for multiple testing throughout the year.

Another popular diagnostic test for PPID is to measure baseline plasma ACTH levels. ACTH is one of the POMC-derived peptides that is over expressed by melanotropes in PPID affected horses [61]. ACTH levels are elevated in PPID affected horses when compared to healthy controls [51, 67, 69]. This test requires a single blood draw [51] for plasma and is tested using a chemiluminescent ACTH immunoassay

9 validated specifically for equine plasma [71]. Reported reference values have varied between 35 pg/mL [28, 51], 50 pg/mL [72] and – 70 pg/mL [15], though the 35 pg/mL cutoff seems to be used most often in clinical practice [22]. Testing for endogenous

ACTH also faces complications due to seasonal fluctuations as both PPID affected and normal horses reach their highest plasma concentrations in the autumn from August to

October [28, 69], though the total percent increase in ACTH concentration is greater in

PPID horses [28]. It has been proposed that a separate reference range for ACTH levels may be established for PPID affected horses in the autumn [69]. ACTH levels may also fluctuate throughout the year in response to stresses such as laminitis [73] and transportation [74]. However, baseline plasma ACTH levels are still advantageous over dexamethasone suppression tests, especially if repeated testing is needed to calibrate medication, because only a single blood draw and, therefore, a single veterinarian visit is required. While a single elevated endogenous ACTH level may be indicative of

PPID, it may be more informative to test throughout the year to make an accurate diagnosis [75].

α-MSH, a POMC-derived peptide, has been investigated as another potential diagnostic marker for PPID [28]. αMSH suffers the same problem as ACTH in that levels are elevated during the autumn in both horses and ponies [76]. α-MSH shows a correlation with body mass index [77] and may not be an accurate diagnostic tool for horses that have lost body condition due to muscle wasting. There may also be a need for separate reference values based on breed as one study has shown an 11-fold seasonal increase in α-MSH in horses as compared to a 2-fold higher level in ponies

[76]. Though it makes intuitive sense to measure cortisol concentrations to diagnose

10 PPID, cortisol levels are variable in PPID horses [48] and therefore does not make a reliable diagnostic test.

Treatments for PPID

Some of the most effective treatments currently available to owners of PPID affected horses are pergolide, cyproheptadine and trilostane [71, 78-80]. Pergolide is a dopamine agonist that acts to replace the action of the damaged dopaminergic neurons of the hypothalamus. Pergolide was evaluated as a treatment for Parkinson’s Disease in humans [81]. In PPID affected horses, it reduces plasma ACTH concentrations [15, 71,

82]. Cyproheptaine, a serotonin antagonist, is less effective in terms of improving clinical signs of PPID when compared with pergolide [82]. A complication associated with pergolide is that dosing is difficult because severity of PPID may not be determined until post mortem examination of the pituitary and, even then, it is still not definitive [83]. Repeated baseline ACTH levels may need to be drawn to help assess the effectiveness of the treatment. It has yet to be determined if pergolide arrests the hyperplasia of the pituitary, though this is likely dose dependent.

Trilostane is a competitive inhibitor of the 3β-hydroxysteroid dehydrogenase.

Widely used as a therapy for hyperadrenocorticism in dogs [84] and shown to be effective in horses [79], it acts to decrease circulating cortisol levels. It may not be the best treatment because, as mentioned above, cortisol levels are not always elevated in

PPID affected horses. Also, trilostane does not control the hyperplasia of the pars intermedia. Pergolide is the better treatment because it likely also restores the balance of all POMC-derived peptides which may account for all of the symptoms where trilostane will only help control any phenotypic changes due to cortisol. PPID affected

11 horses would likely benefit from the same management of diet and exercise as EMS horses to help reduce obesity.

Identifying Genetic Predispostions

Though no specific genetic predispositions for EMS or ECD have been identified, breed differences have been noted in previous studies, with ponies frequently observed as particularly predisposed to both conditions [4, 65, 85-87]. Ponies have higher circulating levels of insulin when compared to horses [85, 88] and there is a notable difference in insulin sensitivity across breeds [88]. There may also be interplay between EMS and PPID, with PPID predisposing to hyperinsulinemia [15, 89, 90].

However Frank et al. has shown that mean insulin levels do not differ between PPID and normal horses and hyperinsulinemia was not observed after overnight stall confinement [69]. EMS horses have the same seasonal ACTH pattern as normal horses

[22] which shows that EMS can develop independent of PPID. However, the overlap of clinical signs between these two conditions and the possible mechanisms that may predispose one condition to another warrants investigation of both conditions for possible genetic predispositions.

REFERENCES

1. Pollitt CC. (1996) Basement membrane pathology: A feature of acute equine laminitis. Equine Vet J 28(1): 38-46.

2. Belknap JK. (2010) Black walnut extract: An inflammatory model. Vet Clin North Am Equine Pract 26(1): 95-101.

3. McGowan CM. (2010) Endocrinopathic laminitis. Vet Clin North Am Equine Pract 26(2): 233-237.

4. Johnson PJ. (2002) The equine metabolic syndrome peripheral cushing's syndrome. Vet Clin North Am Equine Pract 18(2): 271-293.

5. Wild SH, Byrne CD. (2006) ABC of obesity. risk factors for diabetes and coronary heart disease. BMJ 333(7576): 1009-1011.

6. Henneke DR, Potter GD, Kreider JL, Yeates BF. (1983) Relationship between condition score, physical measurements and body fat percentage in mares. Equine Vet J 15(4): 371-372.

7. Hoffman RM, Boston RC, Stefanovski D, Kronfeld DS, Harris PA. (2003) Obesity and diet affect glucose dynamics and insulin sensitivity in thoroughbred geldings. J Anim Sci 81(9): 2333-2342.

8. Frank N, Elliott SB, Brandt LE, Keisler DH. (2006) Physical characteristics, blood hormone concentrations, and plasma lipid concentrations in obese horses with insulin resistance. J Am Vet Med Assoc 228(9): 1383-1390.

9. Bailey SR, Habershon-Butcher JL, Ransom KJ, Elliott J, Menzies-Gow NJ. (2008) Hypertension and insulin resistance in a mixed-breed population of ponies predisposed to laminitis. Am J Vet Res 69(1): 122-129.

10. Alberti KG, Zimmet P, Shaw J. (2006) Metabolic syndrome--a new world-wide definition. A consensus statement from the international diabetes federation. Diabet Med 23(5): 469-480.

11. Carter RA, Geor RJ, Burton Staniar W, Cubitt TA, Harris PA. (2009) Apparent adiposity assessed by standardised scoring systems and morphometric measurements in horses and ponies. Vet J 179(2): 204-210.

12. Vick MM, Adams AA, Murphy BA, Sessions DR, Horohov DW, et al. (2007) Relationships among inflammatory cytokines, obesity, and insulin sensitivity in the horse. J Anim Sci 85(5): 1144-1155.

13 13. Carter RA, McCutcheon LJ, George LA, Smith TL, Frank N, et al. (2009) Effects of diet-induced weight gain on insulin sensitivity and plasma hormone and lipid concentrations in horses. Am J Vet Res 70(10): 1250-1258.

14. Frank N, Geor RJ, Bailey SR, Durham AE, Johnson PJ, et al. (2010) Equine metabolic syndrome. J Vet Intern Med 24(3): 467-475.

15. Walsh DM, McGowan CM, McGowan T, Lamb SV, Schanbacher BJ, et al. (2009) Correlation of plasma insulin concentration with laminitis score in a field study of equine cushing's disease and equine metabolic syndrome. Journal of Equine Veterinary Science 29(2): 87-94.

16. Treiber KH, Kronfeld DS, Hess TM, Byrd BM, Splan RK, et al. (2006) Evaluation of genetic and metabolic predispositions and nutritional risk factors for pasture-associated laminitis in ponies. J Am Vet Med Assoc 228(10): 1538- 1545.

17. Asplin KE, Sillence MN, Pollitt CC, McGowan CM. (2007) Induction of laminitis by prolonged hyperinsulinaemia in clinically normal ponies. Vet J 174(3): 530-535.

18. de Laat MA, McGowan CM, Sillence MN, Pollitt CC. (2010) Equine laminitis: Induced by 48 h hyperinsulinaemia in standardbred horses. Equine Vet J 42(2): 129-135.

19. Weyer C, Tataranni PA, Bogardus C, Pratley RE. (2001) Insulin resistance and insulin secretory dysfunction are independent predictors of worsening of glucose tolerance during each stage of type 2 diabetes development. Diabetes Care 24(1): 89-94.

20. Durham AE, Rendle DI, Newton JE. (2008) The effect of metformin on measurements of insulin sensitivity and beta cell response in 18 horses and ponies with insulin resistance. Equine Vet J 40(5): 493-500.

21. Durham AE, Hughes KJ, Cottle HJ, Rendle DI, Boston RC. (2009) Type 2 diabetes mellitus with pancreatic beta cell dysfunction in 3 horses confirmed with minimal model analysis. Equine Vet J 41(9): 924-929.

22. Place NJ, McGowan CM, Lamb SV, Schanbacher BJ, McGowan T, et al. (2010) Seasonal variation in serum concentrations of selected metabolic hormones in horses. J Vet Intern Med 24(3): 650-654.

23. Suagee JK, Corl BA, Hulver MW, McCutcheon LJ, Geor RJ. (2011) Effects of hyperinsulinemia on glucose and lipid transporter expression in insulin-sensitive horses. Domest Anim Endocrinol 40(3): 173-181.

14 24. Toth F, Frank N, Martin-Jimenez T, Elliott SB, Geor RJ, et al. (2010) Measurement of C-peptide concentrations and responses to somatostatin, glucose infusion, and insulin resistance in horses. Equine Vet J 42(2): 149-155.

25. Duehlmeier R, Hacker A, Widdel-Bigdely A, von Engelhardt W, Sallmann HP. (2010) Insulin stimulates GLUT4 translocation in the semitendinosus muscle of shetland ponies. Vet J 184(2): 176-181.

26. Waller AP, Burns TA, Mudge MC, Belknap JK, Lacombe VA. (2011) Insulin resistance selectively alters cell-surface glucose transporters but not their total protein expression in equine skeletal muscle. Journal of Veterinary Internal Medicine.

27. Gordon ME, McKeever KH. (2005) Diurnal variation of ghrelin, leptin, and adiponectin in standardbred mares. J Anim Sci 83(10): 2365-2371.

28. Beech J, Boston RC, McFarlane D, Lindborg S. (2009) Evaluation of plasma ACTH, alpha-melanocyte-stimulating hormone, and insulin concentrations during various photoperiods in clinically normal horses and ponies and those with pituitary pars intermedia dysfunction. J Am Vet Med Assoc 235(6): 715-722.

29. Carter RA, Treiber KH, Geor RJ, Douglass L, Harris PA. (2009) Prediction of incipient pasture-associated laminitis from hyperinsulinaemia, hyperleptinaemia and generalised and localised obesity in a cohort of ponies. Equine Vet J 41(2): 171-178.

30. Eiler H, Frank N, Andrews FM, Oliver JW, Fecteau KA. (2005) Physiologic assessment of blood glucose homeostasis via combined intravenous glucose and insulin testing in horses. Am J Vet Res 66(9): 1598-1604.

31. Frank N. (2009) Equine metabolic syndrome. Journal of Equine Veterinary Science 29(5): 259-267.

32. Pratt SE, Geor RJ, McCutcheon LJ. (2005) Repeatability of 2 methods for assessment of insulin sensitivity and glucose dynamics in horses. Journal of Veterinary Internal Medicine 19(6): 883-888.

33. Carter RA, McCutcheon LJ, Valle E, Meilahn EN, Geor RJ. (2010) Effects of exercise training on adiposity, insulin sensitivity, and plasma hormone and lipid concentrations in overweight or obese, insulin-resistant horses. Am J Vet Res 71(3): 314-321.

34. Respondek F, Myers K, Smith TL, Wagner A, Geor RJ. (2010) Dietary supplementation with short-chain fructo-oligosaccharides improves insulin sensitivity in obese horses. J Anim Sci .

15 35. Treiber KH, Kronfeld DS, Hess TM, Boston RC, Harris PA. (2005) Use of proxies and reference quintiles obtained from minimal model analysis for determination of insulin sensitivity and pancreatic beta-cell responsiveness in horses. Am J Vet Res 66(12): 2114-2121.

36. Toth F, Frank N, Elliott SB, Perdue K, Geor RJ, et al. (2009) Optimisation of the frequently sampled intravenous glucose tolerance test to reduce urinary glucose spilling in horses. Equine Vet J 41(9): 844-851.

37. Nathan DM, Buse JB, Davidson MB, Heine RJ, Holman RR, et al. (2006) Management of hyperglycemia in type 2 diabetes: A consensus algorithm for the initiation and adjustment of therapy: A consensus statement from the american diabetes association and the european association for the study of diabetes. Diabetes Care 29(8): 1963-1972.

38. Vick MM, Sessions DR, Murphy BA, Kennedy EL, Reedy SE, et al. (2006) Obesity is associated with altered metabolic and reproductive activity in the mare: Effects of metformin on insulin sensitivity and reproductive cyclicity. Reprod Fertil Dev 18(6): 609-617.

39. Hustace JL, Firshman AM, Mata JE. (2009) Pharmacokinetics and bioavailability of metformin in horses. Am J Vet Res 70(5): 665-668.

40. Qiao W, Peng Z, Wang Z, Wei J, Zhou A. (2009) Chromium improves glucose uptake and metabolism through upregulating the mRNA levels of IR, GLUT4, GS, and UCP3 in skeletal muscle cells. Biol Trace Elem Res 131(2): 133-142.

41. Suagee JK, Corl BA, Wearn JG, Crisman MV, Hulver MW, et al. (2011) Effects of the insulin-sensitizing drug pioglitazone and lipopolysaccharide administration on insulin sensitivity in horses. Journal of Veterinary Internal Medicine.

42. Knowler WC, Barrett-Connor E, Fowler SE, Hamman RF, Lachin JM, et al. (2002) Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin. N Engl J Med 346(6): 393-403.

43. Vervuert I, Voigt K, Hollands T, Cuddeford D, Coenen M. (2009) Effect of feeding increasing quantities of starch on glycaemic and insulinaemic responses in healthy horses. Vet J 182(1): 67-72.

44. Frank N, Sojka J, Messer NT. (2002) Equine thyroid dysfunction. Veterinary Clinics of North America: Equine Practice 18(2): 305-319.

45. Lowe JE, Baldwin BH, Foote RH, Hillman RB, Kallfelz FA. (1974) Equine hypothyroidism: The long term effects of thyroidectomy on metabolism and growth in mares and stallions. Cornell Vet 64(2): 276-295.

16 46. Vischer CM, Foreman JH, Constable PD, Benson GJ, Kline KH, et al. (1999) Hemodynamic effects of thyroidectomy in sedentary horses. Am J Vet Res 60(1): 14-21.

47. Toth F, Frank N, Geor RJ, Boston RC. (2010) Effects of pretreatment with dexamethasone or levothyroxine sodium on endotoxin-induced alterations in glucose and insulin dynamics in horses. Am J Vet Res 71(1): 60-68.

48. Dybdal NO, Hargreaves KM, Madigan JE, Gribble DH, Kennedy PC, et al. (1994) Diagnostic testing for pituitary pars intermedia dysfunction in horses. J Am Vet Med Assoc 204(4): 627-632.

49. Schott HC,2nd. (2002) Pituitary pars intermedia dysfunction: Equine cushing's disease. Vet Clin North Am Equine Pract 18(2): 237-270.

50. Moore JN, Steiss J, Nicholson WE, Orth DN. (1979) A case of pituitary adrenocorticotropin-dependent cushing's syndrome in the horse. Endocrinology 104(3): 576-582.

51. Couetil L, Paradis MR, Knoll J. (1996) Plasma adrenocorticotropin concentration in healthy horses and in horses with clinical signs of hyperadrenocorticism. J Vet Intern Med 10(1): 1-6.

52. van der Kolk JH, Heinrichs M, van Amerongen JD, Stooker RC, in de Wal LJ, et al. (2004) Evaluation of pituitary gland anatomy and histopathologic findings in clinically normal horses and horses and ponies with pituitary pars intermedia adenoma. Am J Vet Res 65(12): 1701-1707.

53. McFarlane D, Dybdal N, Donaldson MT, Miller L, Cribb AE. (2005) Nitration and increased alpha-synuclein expression associated with dopaminergic neurodegeneration in equine pituitary pars intermedia dysfunction. J Neuroendocrinol 17(2): 73-80.

54. Heinrichs M, Baumgartner W, Capen CC. (1990) Immunocytochemical demonstration of proopiomelanocortin-derived peptides in pituitary adenomas of the pars intermedia in horses. Vet Pathol 27(6): 419-425.

55. Glover CM, Miller LM, Dybdal NO, Lopez A, Duckett WM, et al. (2009) Extrapituitary and pituitary pathological findings in horses with pituitary pars intermedia dysfunction: A retrospective study. Journal of Equine Veterinary Science 29(3): 146-153.

56. Loeb WF, Capen CC, Johnson LE. (1966) Adenomas of the pars intermedia associated with hyperglycemia and glycosuria in two horses. Cornell Vet 56(4): 623-639.

17 57. Orth DN, Nicholson WE. (1982) Bioactive and immunoreactive adrenocorticotropin in normal equine pituitary and in pituitary tumors of horses with cushing's disease. Endocrinology 111(2): 559-563.

58. Millington WR, Dybdal NO, Dawson R,Jr, Manzini C, Mueller GP. (1988) Equine cushing's disease: Differential regulation of beta-endorphin processing in tumors of the intermediate pituitary. Endocrinology 123(3): 1598-1604.

59. van der Kolk JH, Ijzer J, Overgaauw PA, van der Linde-Sipman JS. (2001) Pituitary-independent cushing's syndrome in a horse. Equine Vet J 33(1): 110- 112.

60. Saland LC. (2001) The mammalian pituitary intermediate lobe: An update on innervation and regulation. Brain Res Bull 54(6): 587-593.

61. Saiardi A, Borrelli E. (1998) Absence of dopaminergic control on melanotrophs leads to cushing's-like syndrome in mice. Mol Endocrinol 12(8): 1133-1139.

62. Beyer HS, Matta SG, Sharp BM. (1988) Regulation of the messenger ribonucleic acid for corticotropin-releasing factor in the paraventricular nucleus and other brain sites of the rat. Endocrinology 123(4): 2117-2123.

63. Love S. (1993) Equine cushing's disease. Br Vet J 149(2): 139-153.

64. van der Kolk JH, Kalsbeek HC, van Garderen E, Wensing T, Breukink HJ. (1993) Equine pituitary neoplasia: A clinical report of 21 cases (1990-1992). Vet Rec 133(24): 594-597.

65. HILLYER MH, TAYLOR FGR, MAIR TS, MURPHY D, WATSON TDG, et al. (1992) Diagnosis of hyperadrenocorticism in the horse. Equine Veterinary Education 4(3): 131-134.

66. Frank N, Andrews FM, Sommardahl CS, Eiler H, Rohrbach BW, et al. (2006) Evaluation of the combined dexamethasone suppression/ thyrotropin-releasing hormone stimulation test for detection of pars intermedia pituitary adenomas in horses. J Vet Intern Med 20(4): 987-993.

67. van der Kolk JH, Wensing T, Kalsbeek HC, Breukink HJ. (1995) Laboratory diagnosis of equine pituitary pars intermedia adenoma. Domest Anim Endocrinol 12(1): 35-39.

68. Donaldson MT, McDonnell SM, Schanbacher BJ, Lamb SV, McFarlane D, et al. (2005) Variation in plasma adrenocorticotropic hormone concentration and dexamethasone suppression test results with season, age, and sex in healthy ponies and horses. J Vet Intern Med 19(2): 217-222.

18 69. Frank N, Elliott SB, Chameroy KA, Toth F, Chumbler NS, et al. (2010) Association of season and pasture grazing with blood hormone and metabolite concentrations in horses with presumed pituitary pars intermedia dysfunction. J Vet Intern Med 24(5): 1167-1175.

70. Lincoln GA, Rhind SM, Pompolo S, Clarke IJ. (2001) Hypothalamic control of photoperiod-induced cycles in food intake, body weight, and metabolic hormones in rams. Am J Physiol Regul Integr Comp Physiol 281(1): R76-90.

71. Perkins GA, Lamb S, Erb HN, Schanbacher B, Nydam DV, et al. (2002) Plasma adrenocorticotropin (ACTH) concentrations and clinical response in horses treated for equine cushing's disease with cyproheptadine or pergolide. Equine Vet J 34(7): 679-685.

72. Haritou SJ, Zylstra R, Ralli C, Turner S, Tortonese DJ. (2008) Seasonal changes in circadian peripheral plasma concentrations of melatonin, serotonin, dopamine and cortisol in aged horses with cushing's disease under natural photoperiod. J Neuroendocrinol 20(8): 988-996.

73. Ayala I, Martos NF, Silvan G, Gutierrez-Panizo C, Clavel JG, et al. Cortisol, adrenocorticotropic hormone, serotonin, adrenaline and noradrenaline serum concentrations in relation to disease and stress in the horse. Res Vet Sci.

74. Fazio E, Medica P, Aronica V, Grasso L, Ferlazzo A. (2008) Circulating beta- endorphin, adrenocorticotrophic hormone and cortisol levels of stallions before and after short road transport: Stress effect of different distances. Acta Vet Scand 50: 6.

75. Lee ZY, Zylstra R, Haritou SJ. (2010) The use of adrenocorticotrophic hormone as a potential biomarker of pituitary pars intermedia dysfunction in horses. Vet J 185(1): 58-61.

76. McFarlane D, Donaldson MT, McDonnell SM, Cribb AE. (2004) Effects of season and sample handling on measurement of plasma alpha-melanocyte- stimulating hormone concentrations in horses and ponies. Am J Vet Res 65(11): 1463-1468.

77. Donaldson MT, McFarlane D, Jorgensen AJ, Beech J. (2004) Correlation between plasma alpha-melanocyte-stimulating hormone concentration and body mass index in healthy horses. Am J Vet Res 65(11): 1469-1473.

78. Donaldson MT, Jorgensen AJ, Beech J. (2004) Evaluation of suspected pituitary pars intermedia dysfunction in horses with laminitis. J Am Vet Med Assoc 224(7): 1123-1127.

19 79. McGowan CM, Neiger R. (2003) Efficacy of trilostane for the treatment of equine cushing's syndrome. Equine Vet J 35(4): 414-418.

80. Gehring R, Beard L, Wright A, Coetzee J, Havel J, et al. (2010) Single-dose oral pharmacokinetics of pergolide mesylate in healthy adult mares. Vet Ther 11(1): E1-8.

81. Arnold G, Gasser T, Storch A, Lipp A, Kupsch A, et al. (2005) High doses of pergolide improve clinical global impression in advanced parkinson's disease:- a preliminary open label study. Arch Gerontol Geriatr 41(3): 239-253.

82. Donaldson MT, LaMonte BH, Morresey P, Smith G, Beech J. (2002) Treatment with pergolide or cyproheptadine of pituitary pars intermedia dysfunction (equine cushing's disease). J Vet Intern Med 16(6): 742-746.

83. McFarlane D, Miller LM, Craig LE, Dybdal NO, Habecker PL, et al. (2005) Agreement in histologic assessments of the pituitary pars intermedia in aged horses. Am J Vet Res 66(12): 2055-2059.

84. Galac S, Buijtels JJ, Mol JA, Kooistra HS. (2010) Effects of trilostane on the pituitary-adrenocortical and renin-aldosterone axis in dogs with pituitary- dependent hypercortisolism. Vet J 183(1): 75-80.

85. Jeffcott LB, Field JR, McLean JG, O'Dea K. (1986) Glucose tolerance and insulin sensitivity in ponies and standardbred horses. Equine Vet J 18(2): 97-101.

86. Coffman JR, Colles CM. (1983) Insulin tolerance in laminitic ponies. Can J Comp Med 47(3): 347-351.

87. Karikoski NP, Horn I, McGowan TW, McGowan CM. (2011) The prevalence of endocrinopathic laminitis among horses presented for laminitis at a first- opinion/referral equine hospital. Domest Anim Endocrinol .

88. Firshman AM, Valberg SJ. (2007) Factors affecting clinical assessment of insulin sensitivity in horses. Equine Vet J 39(6): 567-575.

89. Garcia MC, Beech J. (1986) Equine intravenous glucose tolerance test: Glucose and insulin responses of healthy horses fed grain or hay and of horses with pituitary adenoma. Am J Vet Res 47(3): 570-572.

90. McGowan CM, Frost R, Pfeiffer DU, Neiger R. (2004) Serum insulin concentrations in horses with equine cushing's syndrome: Response to a cortisol inhibitor and prognostic value. Equine Vet J 36(3): 295-298.

20 CHAPTER 2 Original research Background

The two most common conditions leading to endocrinopathic laminitis in the horse are equine metabolic syndrome (EMS) and equine Cushing’s Disease (ECD), also known as pituitary pars intermedia dysfunction (PPID) [1]. Both of these conditions present similar clinical signs such as regional adiposity. It has been suggested that one condition may predispose to the other as high plasma concentrations of both insulin and ACTH in the same individual are frequently observed. Though no specific genetic predispositions to either condition have been identified, a difference in the prevalence of either condition across breeds suggests a genetic link.

PPID results from the loss of dopaminergic inhibition from the neurons of the periventricular nucleus [2] leading to increased production of proopiomelanocortin and, subsequently, ACTH. The mean age of onset for PPID is 20 years [3, 4] but cases have been observed in individuals as young as 7 years old [5]. An abnormal hair coat is the most obvious clinical sign of PPID [4] and is not observed in EMS affected horses.

When diagnosing PPID the dexamethasone suppression test is considered the gold standard, though it is not reliable during the seasonal ACTH increase of autumn [6].

Endogenous plasma ACTH levels are preferable to the dexamethasone suppression test because only one plasma sample is needed, but it also is not a reliable measurement during the autumn [7, 8].

The key clinical signs of EMS (previously referred to as Peripheral Cushing’s

Disease) are regional adiposity, often presenting as a “cresty” neck, insulin resistance

21 characterized by hyperinsulinemia and a history of laminitis [9]. The average age of onset for EMS horses is slightly younger than that of PPID with ages often ranging from 8-18 years [9]. Diagnostic testing available for EMS include labor intensive methods such as the frequently sampled intravenous glucose tolerance test, the combined glucose insulin test and the euglycemic-hyperinsulinemic clamp technique.

For clinical applications, it is simplest to draw a single blood sample and measure the fasted plasma insulin concentration to detect hyperinsulinemia.

Currently, there are few pharmacological options available to treat PPID or

EMS affected horses. PPID is often successfully managed using the dopamine agonist pergolide, but it is usually given daily and dosage varies from 0.5 mg to 5 mg. At these dosages pergolide is an expensive option. Cyproheptadine, a serotonin antagonist, and trilostane, an inhibitor of β-hydroxysteroid dehydrogenase, are alternative treatments, but pergolide remains the most effective medication available.

EMS horses have very limited pharmacological treatment options when considering insulin-sensitizing drugs. Responses to metformin treatment in EMS affected individuals have been mixed, possibly due to limited bioavailability of this drug in the horse.

The interaction between EMS and PPID has yet to be determined. Although

PPID affected horses are often observed to also have hyperinsulinemia [10-12], Frank et al. showed that mean insulin levels do not differ between PPID and control horses and hyperinsulinemia was not observed after overnight stall confinement [13]. In addition, horses are able to maintain normal ACTH concentrations and seasonal patterns in the presence of hyperinsulinemia [7]. However, the frequent observation of concurrent

22 elevated plasma concentrations in both insulin and ACTH indicates that these conditions may predispose an individual to develop the other condition.

The differences observed in the frequency of either PPID or EMS across breeds indicates that there may be a genetic component to either condition. Treiber et al. suggest a dominant mode of inheritance for laminitis in individuals with EMS [14], though the research herd for that study consisted exclusively of ponies and the methods of pedigree analysis were not made available. Differences in the rate of elevated plasma concentrations of ACTH, an indication of PPID, and insulin, an indication of EMS, across different breeds shows that ponies, Morgans and Arabians have the highest frequencies of hyperinsulinemia while Thoroughbreds and Quarter

Horses rank amongst the lowest frequencies (Streeter et al., unpublished data). This is consistent with previous observations that ponies and Morgans are at a higher risk for

EMS [15]. Arabians were chosen as the target breed for this study due to their moderate frequency of EMS and PPID as well as their large population and, consequently, ease of obtaining DNA samples in sufficient quantity for the study. Furthermore, any candidate genes identified in this study can be investigated in Arabian-derived breeds with low disease frequencies, such as the Quarter Horse and Thoroughbred that may have lost genetic predispositions to both conditions.

The objective of this study is to conduct a genome-wide association to identify candidate genes associated with EMS, PPID or both conditions within the Arabian breed using the EquineSNP50 chip. The EquineSNP50 chip has been used successfully to identify candidate genes for single gene diseases [16] as well as likely polygenic traits [17].

23 Materials and Methods Horses – Horses were recruited for this study based on endocrinology records from Cornell University College of Veterinary Medicine’s Animal Health Diagnostic

Center. Owners were contacted and asked to submit a brief health history accompanied by a blood or hair sample. A total of sixty-five samples from Arabian horses were obtained for this study. All but five horses were confirmed to be of full Arabian ancestry based on registration records. Of the five individuals that were not confirmed to be full Arabians, all were anecdotally reported to be of full or a majority of Arabian descent although registration papers were not available. Ages, based on known ages at the time of the blood draw, ranged from 11 to 30 years. The mean ages for the disease categories Both, EMS, PPID, Normal and were 25.0, 18.6, 22.8, and 19.4 years, respectively. The group consisted of 32 geldings, 30 mares and 3 stallions.

Classification – Classification of horses was adapted from Walsh et al. into PPID (P), EMS (M), Normal (N) and an additional Both (B) group [11]. Horses with ACTH levels greater than 70 pg/mL and insulin levels less than 40 uIU/mL were designated “PPID” while horses with insulin levels > 70 uIU/mL and ACTH levels

<40 pg/mL were designated “EMS”. Horses with ACTH and insulin levels <40 pg/mL or uIU/mL, respectively, were considered “Normal” and those with both levels > 70 were grouped into the fourth category termed “Both”. Individuals with levels between

40 and 70 were omitted as uncertainty in reference ranges made it difficult to assign those individuals to a disease category. Horses currently on medications were evaluated based on endocrinology results prior to starting medication. Horses with elevated levels for both insulin and ACTH were assigned to either the EMS or PPID groups if previous endocrinology testing was available to indicate which condition developed first. Levels taken between the months of August and October were not used to determine disease status due to the seasonal increase in plasma ACTH levels

24 [7]. The total number of individuals in each group for Both, EMS, PPID and Normal were 9, 15, 24 and 17, respectively. DNA extraction and genotyping – Blood samples were extracted using the

Gentra Puregene Blood Kit (Qiagen Inc, Valencia, CA, USA) using the manufacturer’s protocol for whole blood. Hair samples were extracted using an adapted whole blood protocol starting with 15 hair bulbs incubated overnight at 65oC in 100 µL of lysis buffer [1% Tween 20, 1X PCR buffer (Applied Biosystems, Foster

City, CA, USA), 2.5mM MgCl2] containing 2 µg Proteinase K. The extraction was completed following the manufacturer’s protocol. A total of 500 ng of each DNA sample was sent for SNP genotyping at GeneSeek (Lincoln, NE, USA) where Illumina’s Infinium® II Assay (Illumina Inc, San Diego, CA, USA) was performed on the EquineSNP 50 Genotyping BeadChip (Illumina Inc, San Diego, CA, USA) containing 54,602 SNPs. The EquineSNP 50 BeadChip SNPs were derived from the EquCab2.0 assembly with an average spacing of 43.2 kb across the genome. Statistical Analysis – Statistical analyses were performed using the PLINK v1.07 Whole genome association analysis toolset [18] and JMP 8.0 (SAS Insitute Inc,

Cary, NC, USA). Basic association (chi-square) tests were used to compare disease cases and controls. Since it is unknown whether there are common genetic predisposing factors between EMS and PPID or if they develop independently, various groupings of the disease and control categories were used to develop a total of 15 qualitative associations. Quantitative associations were also performed for highest recorded ACTH (excluding August-October measurements) and insulin levels. An additional association was performed comparing horses with at least one episode of laminitis versus those with no recorded history of laminitis, regardless of disease category. SNPs with a missing genotype rate greater than 10% were excluded from analysis. SNPs in linkage disequilibrium (r2 > 0.99) were pruned to a single SNP per

25 haplotype block in all graphs. P-values unadjusted for multiple testing are reported and the Bonferroni corrected α level is shown where appropriate. Overall study α level is 5%. Any SNPs exceeding the significance threshold for multiple testing were mapped on the UCSC Genome Browser [19] using the September 2007 assembly (EquCab2.0) of the horse genome. Candidate genes were identified within ~50 kb of each significant SNP.

Results

A total of 68 SNPs representing 59 different loci exceeded the significance threshold after correction for multiple testing. When performing a quantitative association with the highest ACTH level recorded for each individual, SNP BIEC2-

215377 (ECA 13) was the most significant SNP with a P-value of 4.34 e-7 (Figure

2.1). Two additional significant SNPs, BIEC2-377685 (ECA17) and BIEC2-1007280

(ECA 7) with P-values of 3.55 e-6 and 2.4 e-5, respectively, were also found. XPO6 was the only candidate gene identified within 50 kb of SNP BIEC2-215377 (ECA13).

Candidate genes identified for BIEC-1007280 (ECA7) include OLFML1, SYT9 and

PPFIBP2. No candidate genes were identified for BIEC2-377685 (ECA17).

Figure 2.1 Quantitative association of highest recorded ACTH value. Association of 38,315 independent SNPS with –log10P-value on the y-axis and SNP, sorted and color- coded by chromosome, on the x-axis. Reference line included to show the significance threshold for multiple testing. Three significant SNPs are shown, including the most significant SNP, SNP BIEC2-215377 (ECA13) with a P-value of 4.34 e-7.

The SNP BIEC2-173131 (ECA12) achieved the second best P-value (1.48 e-6) of all associations in the quantitative association of the highest recorded insulin. This association also gave the greatest number of significant SNPs for a single association with a total of 38 SNPs representing 30 different loci (Figure 2.2). Candidate genes found for BIEC2-173131 include ALX4 and PRRX2.

Figure 2.2 Quantitative association of highest recorded insulin value. Association of 38,343 independent SNPS with –log10P-value on the y-axis and SNP, sorted and color- coded by chromosome, on the x-axis. Reference line included to show the significance threshold for multiple testing. A total of 38 SNPs representing 30 loci exceeded the significance threshold.

Manhattan plots for each association and a table of all significant SNPs can be found in the Appendix. Any SNPs that do not exceed the significance threshold but deviate from expected P-values, as demonstrated by a quantile-quantile plot, may be worth further investigation. Also of particular interest are those SNPs with candidate genes that have been previously associated with obesity, diabetes and/or insulin resistance. A Manhattan plot of BM vs. N shows only one significant SNP, BIEC2-

596175 (ECA22) (Figure 2.3). The SNP with the second smallest P-value, BIEC2-

770354 (ECA3) approaches but does not exceed the significance threshold for multiple testing.

Figure 2.3 Qualitative association of hyperinsulinemia (B or M) vs. controls (N). Association of 37,647 independent SNPS with –log10P-value on the y-axis and SNP, sorted and color-coded by chromosome, on the x-axis. Reference line included to show the significance threshold for multiple testing.

Two candidate genes, FTO and ATP5H, were identified for BIEC2-770354

(ECA3) and its neighboring SNP in linkage disequilibrium, BIEC2-770355. Both

SNPs are located within 500 bp of each other and are found in the intron of FTO in the human, mouse and cow. These two SNPs also deviate from expected –log10 P-values in the association BMP vs. N with a P-value of 1.03 e-4 (Figure 2.4).

Figure 2.4 Quantile-quantile plot of qualitative association of all disease groups (B, M and P) vs. controls (N). The actual –log10 P-value is plotted on the y-axis with the expected –log10 P-value plotted on the x-axis.

Another candidate gene, attractin-like 1 (ATRNL1), was found in the association M vs. N. Two SNPs, BIEC2-7588 (ECA1) and BIEC2-7581 (ECA1), located within 7 kb of the first intron of ATRNL1 did not exceed the significance threshold but did deviate from expected –log10 P-value with P-values of 1.01 e-4 and.

1.26 e-4, respectively.

Discussion

The lack of affordable and effective treatments for both EMS and PPID combined with the evidence of possible genetic predispositions to both conditions warrants further investigation into the heritable components of both disorders.

Identification of the genetic underpinnings of EMS and PPID allows for a better understanding of how both conditions arise and if alternative therapies can be

30 developed. This genome-wide association study has identified several interesting candidate genes previously linked to obesity, diabetes and/or insulin resistance in humans and other animals. The most intriguing candidate is the fat mass and obesity associated gene or FTO. Fto was first identified in the rat by Peters et al. who designated it “fatso” due to its large, 250 kb sequence length [20]. This proved to be very serendipitous naming as FTO was first recognized as a predisposing factor to obesity in a genome-wide association study looking for genes associated with type II diabetes in humans [21]. Frayling et al. found that SNPs within the first intron of FTO were associated with type II diabetes through an effect on body mass index [21]. Since that original paper FTO variants have been associated with obesity in a wide variety of populations [22-24]. Though the function of FTO is not known, there have been many studies attempting to elucidate the mechanism of how FTO predisposes individuals to obesity. FTO expression is increased in the skeletal muscle of type II diabetic patients

[25] as well as in the subcutaneous adipose tissue of obese individuals [26]. Murine Fto has been shown to increase feed intake and, therefore, increase obesity regardless of diet type [27]. Furthermore, this same study showed that a lack of Fto function results in a lean phenotype in mice [27].

FTO is an excellent candidate gene to investigate in both EMS and PPID affected horses due to the link with body mass index. The gene is highly conserved across all vertebrate species including teleost fish suggesting a history in vertebrate evolution of over 450 million years [28]. It is also worthy to note that, in the rat, Fto expression is detected in the paraventricular nucleus and is up regulated in the hypothalamus in response to 48 hours of food deprivation [28]. The exact mechanism

31 by which FTO regulates feed intake is unknown; therefore, if FTO is capable of regulating feed intake in PPID affected horses in the face of degeneration of the dopaminergic neurons of the paraventricular nucleus this may suggest a pathway independent of the innervation of the pars intermedia. FTO shows its best P-value in the association where all metabolic conditions are considered (p = 4.16 e-5, ECA3

6.5Mb). As the Both and EMS categories share this common locus, this suggests a genetic predisposition to EMS followed by PPID.

The SNP with the second most significant P-value, BIEC2-173131, yielded two candidate genes, ALX4 and PRRX2. In a genome wide association study to identify loci associated with type II diabetes in humans, Sladek et al. identified ALX4 as a candidate gene [29]. This study used stringent selection criteria to enrich for risk alleles associated with type II diabetes. The first cohort required individuals to have at least one affected first degree relative with type II diabetes. Individuals also had a lower body mass index to help select for alleles that predispose to insulin resistance through a pathway independent of obesity. A second cohort with relaxed inclusion criteria were then tested to confirm significant loci identified in the first stage. Though no SNPs were identified within ALX4, three SNPs were found within EXT2 in the same haplotype block. ALX4 is an interesting candidate gene as it was found independent of excessive body mass index in the first cohort. Due to insufficient information concerning weight and body measurements, it would be difficult to assess the relation of ALX4 to hyperinsulinemia independent of obesity in the present study.

However investigation of ALX4 in an experimental herd of hyperinsulinemic and

32 control horses with varying levels of obesity may show an association similar to that found in humans.

Another top candidate gene from this dataset (p = 1.01 e-4) is attractin-like 1 or ATRNL-1. Attractin, a transmembrane protein product of the Mahogany gene [30], is believed to act as a co-receptor with the melanocortin 1 receptor (MC1R) for agouti,

[31]. Attractin-like protein (ALP), unlike attractin, is able to interact directly with the

MC4R C-terminus [32]. ALP has high expression in brain, kidney, heart lung and liver, overlapping with some of the same brain sites where MC4R mRNA is expressed [32]. It has been suggested that attractin has a protective role in the face of oxidative stress in the rat [33] and that attractin and ALP may have redundant roles [34]. This evidence makes a case for a possible role for ATRNL-1 in protection of individuals from the oxidative stress associated with obesity and, in consequence, protection from the development of the neuronal degeneration associated with PPID. Notably, ALP null mice do not seem to show any abnormalities due to the possible redundancy with attractin [34], so any SNPs associated with ATRNL-1 that result in loss-of-function would likely not result in obesity. However, if a SNP within ATRNL-1 were associated with an altered form of ALP that antagonizes MC4R, then obesity may result.

MC4R has been extensively researched for its role in obesity in a variety of animals. MC4R normally acts to regulate energy homeostasis and food intake.

Endogenous agonists for MC4R include the melanocyte-stimulating hormones derived from POMC, with β-MSH having the highest affinity [35]. Agouti-related protein

(AgRP), an antagonist of MC4R, induces increased food intake [36] and prolonged antagonism of MC4R results in hyperphagia and obesity in rats [37]. MC4R has also

33 been associated with elevated insulin and glucose levels [38]. However, the obesity and metabolic syndrome associated with MC4R dysfunction was shown to be overcome through exercise in MC4R knockout mice [39]. This suggests that mutations of

ATRNL-1 that affect its interaction with MC4R may possibly be overcome through environmental control rather than pharmacological intervention. MC4R itself failed to be significantly associated with either EMS or PPID. POMC would also be a likely candidate gene based on its association with PPID and MC4R mutations, but its closest SNP is located over 30 kb away and therefore may not be sufficiently covered by this panel of markers.

The results of this study are very encouraging with several quality candidate genes. However, further refinement of this study is needed to identify the strongest candidates. A major problem confronted by this study is the relatively limited coverage of SNPs across the horse genome. Comparable SNP chips for cows and humans represent over 500,000 and 1 million loci, respectively. As previously mentioned, several top candidate genes based on current literature, such as POMC, coverage was too sparse at those loci and significant associations in these genes may have been missed. This problem can only be overcome with the development of improved SNP chips with better coverage.

Another problem inherent in genome-wide association studies is that complex traits can be the result of many subtle genetic alterations that must come together to develop the disease. Individual alterations may confer a low to moderate risk of developing the disease, but it is the cumulative effect of these alterations that will bring about the phenotype of interest. A basic genome-wide association study,

34 however, is based on generating P-values for individual SNPs, so SNPs associated with low to moderate risk factors may not be distributed in the population in such a way that significance is detected. It is more likely that an individual that is considered unaffected will carry a mutation that confers a low to moderate effect if they do not carry other risk factors, therefore, the chi square statistic for that particular SNP will not be as extreme and significance will not be detected. Alternatively, a SNP conferring low to moderate risk may have a higher frequency in the diseased group and may exceed the significance threshold. However, this SNP may still need the cumulative effect of other risk factors to bring about the disease and, therefore, its predictive power will be overestimated. Complex statistical methods for deciphering patterns of epistasis are still in development and may clarify these relationships in the future.

A vital part of any genome-wide association study is the correct classification of the affected/disease cases versus the unaffected/control cases. Rare, monogenic diseases may be easier to assign a status to, as in the case of Lavender Foal Syndrome

[16], but likely polygenic traits, such as EMS and PPID where several genes are interacting, are more difficult to properly assign to a category. Some of this difficulty lies in the ability to acquire complete health histories and uniform diagnostic testing.

For example, the form of laminitis typically associated with both EMS and PPID is insidious and may not be recognized until postmortem examination. Some owners may not have noted a history of laminitis when in fact the horse was suffering from low grade laminitis without any clinical signs. Therefore, a lack of significant hits within the laminitis association is likely due to miscategorization of unaffected

35 individuals. This problem of incorrect assignment to a category may be overcome by increasing the number of horses in the study. In the case of Lavender Foal Syndrome, a monogenic disease, 36 individuals (6 affected and 30 unaffected) were needed to generate 14 significant SNPs. A complex trait with low to moderate risk factors will likely need more than a total of 65 samples in order to produce many significant hits.

Adding more individuals will help to dilute the effect of incorrectly assigning a few individuals in the overall study.

Further complicating the matter of correctly assigning individuals to disease categories is the influence of environmental factors such as diet and exercise.

Individuals with genetic predispositions that put them at a low risk of developing EMS may develop the disease if they are put on a diet high in nonstructural carbohydrates.

Diets high in nonstructural carbohydrates can lead to repeated bouts of postprandial hyperinsulinemia and, subsequently, insulin resistance [40, 41]. While general dietary information was provided for the individuals in this study, estimation of total nonstructural carbohydrates intake was difficult to determine. Some individuals had switched diets after being diagnosed with EMS and dietary information leading up to diagnosis was not available. Many individuals in this study were not routinely exercised and were kept on pasture at all times. A lack of physical activity can put individuals with low genetic predispositions at a higher risk for developing insulin resistance [42]. Another factor affecting miscategorization is the age of onset, especially for PPID. With the mean age of onset for PPID being 20 years, it may not be appropriate to use younger horses as controls as they may still yet develop the disease.

The mean age of the control group in this study was within the mean age range of the

36 disease categories, but older controls may be more appropriate to use to minimize miscategorization. In addition, some EMS horses may go on to develop PPID and be assigned to the Both category. However, it is more beneficial to identify the first condition that develops to help separate genetic predispositions to one disease or the other. Any common genetic predispositions between EMS and PPID can be found by combining the two groups as one disease category in basic associations.

It is also important to assess the practical applications of this study for the common horse owner. While the mechanisms of the development of both PPID and

EMS are of interest to researchers, the value for horse owners will be derived from new preventative management practices. As quoted from the ACVIM Consensus

Statement “EMS is a complex disorder for which there are more questions than answers at present” [43]. As information becomes available to researchers about how to properly characterize EMS and PPID, these developments need to be transferred to owners to help identify more affected individuals and increase awareness of both disorders.

The prevalence of obesity in horses is difficult to estimate as studies conducted through owner-based reporting show bias towards lower obesity scores [44].

Stephenson et al. found a prevalence of obesity in leisure horses in the UK of 20.6%, which is less than the 45% reported in a study of horses in Scotland [45]. This same study also found that when 15 randomly selected horses from the study were visited by researchers trained in assessing body condition scores, the mean score assigned by researcher was significantly higher than that assigned by the owner. Further, eight owners assigned a body condition score a full point below that of the researcher. When

37 this error rate was extrapolated to the questionnaire data the percentage of obesity more than doubled to 54.1%. It may be that owners are not properly trained to assign body condition scores or, more likely, owners have misconceptions of what constitutes a “healthy” weight for a horse. The muscle wasting seen in PPID affected horses can be confused with the loss of body condition sometimes seen in older horses [46]. In an attempt to avoid weight loss in geriatric horses owners will supplement the diet of their horse above their nutritional needs which may result in obesity and, consequently, insulin resistance and laminitis. It is necessary to educate owners on how to properly assess and maintain an appropriate body condition, especially in the geriatric horse that is at a higher risk of EMS and PPID.

Early detection is very important to avoid laminitis associated with both PPID and EMS. However, early detection is complicated by the fact that clinical signs of both EMS and PPID may be seen much later than the actual onset of either condition.

Older horses show decreased sensitivity to the dexamethasone suppression test in the absence of other clinical sign of PPID [6]. Therefore, it can be difficult to determine the precise onset of PPID in the geriatric horse. It is unknown what extent of pituitary hyperplasia or duration of endogenous ACTH elevation is required to bring about clinical signs of PPID as the mechanisms of most of these signs remain unknown.

There is considerable overlap of clinical and histological findings between PPID affected and healthy aged controls [47, 48]which makes correct diagnosis of PPID, especially early on in the disease, difficult for clinicians.

The identification of genetic predispositions to both EMS and PPID may allow horses to be used as models for the development of metabolic syndrome and

38 Parkinson’s disease in humans. Parkinson’s disease results from the degeneration of dopaminergic neurons of the substantia nigra [49]. Pergolide, the most effective treatment available for PPID, is still being investigated as a treatment for Parkinson’s disease [50]. The horse offers a model for spontaneous dopaminergic neurodegeneration with PPID. By identifying the genetic predispositions associated with PPID in horses, researchers may be able to look for these same predispositions in humans and elucidate the subtle pathways that lead to Parkinson’s in humans.

PPID is also very important to study because it cannot be cured at present.

Pergolide is an effective pharmacological intervention, but it has not yet been shown to shrink or even arrest the hyperplasia of the pars intermedia. Medication must be administered for the rest of the horse’s life. Surgery is not a practical solution for most horse owners, so control of environmental factors or being able to breed the predisposition out of the horse would be more effective. The late onset of PPID also means that owners are less likely to invest in costly solutions unless that particular horse still has value for breeding or performance.

The horse also poses as an interesting model for metabolic syndrome in humans as horses normally lack hyperglycemia in the presence of hyperinsulinemia.

There may be genetic protective measures in the horse that prevents failure of the pancreatic β cells or it may simply be due to the horse’s shortened lifespan relative to humans. The characteristic regional adiposity that confers a higher risk factor for developing laminitis is similar to the accumulation of abdominal fat seen in humans developing metabolic syndrome. Again, the ability to selectively breed against predispositions in the horse can be beneficial for breeders, but EMS may be controlled

39 by environmental factors. Although insulin resistance may be reversible through intense management, it is difficult to implement a program on previously laminitic individuals. The identification of genetic predispositions will allow owners to identify high risk individuals at birth and manage their diet and exercise regimen appropriately before the onset of disease.

The identification of genetic predispositions to EMS and/or PPID is a necessary step toward developing better diagnostic tests for at risk individuals and better therapies for those affected. The candidate genes identified in this study, especially those previously associated with obesity and insulin resistance, are worthy of further fine mapping. These candidate genes should be investigated in both

Arabians and Arabian-derived breeds with low disease frequencies, such as the

Thoroughbred and Quarter Horse. The addition of more individuals, both affected and older controls, will also help strengthen the power of the associations and identify more candidate genes. The dataset can also be improved with expanded marker panels

(SNP chips) with better coverage across the genome. This study is limited by the information currently available concerning diagnosis and pathophysiology of EMS and PPID. The development of more diagnostic tests for other parameters will help classify affected individuals into their appropriate disease category. Finally, identification of genetic predispositions to EMS and PPID will allow owners to breed these predispositions out of their herds and provide better treatment options to those individuals already affected by either condition.

40 REFERENCES

1. McGowan CM. (2010) Endocrinopathic laminitis. Vet Clin North Am Equine Pract 26(2): 233-237.

2. McFarlane D, Dybdal N, Donaldson MT, Miller L, Cribb AE. (2005) Nitration and increased alpha-synuclein expression associated with dopaminergic neurodegeneration in equine pituitary pars intermedia dysfunction. J Neuroendocrinol 17(2): 73-80.

3. van der Kolk JH, Wensing T, Kalsbeek HC, Breukink HJ. (1995) Laboratory diagnosis of equine pituitary pars intermedia adenoma. Domest Anim Endocrinol 12(1): 35-39.

4. HILLYER MH, TAYLOR FGR, MAIR TS, MURPHY D, WATSON TDG, et al. (1992) Diagnosis of hyperadrenocorticism in the horse. Equine Veterinary Education 4(3): 131- 134.

5. Heinrichs M, Baumgartner W, Capen CC. (1990) Immunocytochemical demonstration of proopiomelanocortin-derived peptides in pituitary adenomas of the pars intermedia in horses. Vet Pathol 27(6): 419-425.

6. Donaldson MT, McDonnell SM, Schanbacher BJ, Lamb SV, McFarlane D, et al. (2005) Variation in plasma adrenocorticotropic hormone concentration and dexamethasone suppression test results with season, age, and sex in healthy ponies and horses. J Vet Intern Med 19(2): 217-222.

7. Place NJ, McGowan CM, Lamb SV, Schanbacher BJ, McGowan T, et al. (2010) Seasonal variation in serum concentrations of selected metabolic hormones in horses. J Vet Intern Med 24(3): 650-654.

8. Lee ZY, Zylstra R, Haritou SJ. (2010) The use of adrenocorticotrophic hormone as a potential biomarker of pituitary pars intermedia dysfunction in horses. Vet J 185(1): 58-61.

9. Johnson PJ. (2002) The equine metabolic syndrome peripheral cushing's syndrome. Vet Clin North Am Equine Pract 18(2): 271-293.

10. Garcia MC, Beech J. (1986) Equine intravenous glucose tolerance test: Glucose and insulin responses of healthy horses fed grain or hay and of horses with pituitary adenoma. Am J Vet Res 47(3): 570-572.

11. Walsh DM, McGowan CM, McGowan T, Lamb SV, Schanbacher BJ, et al. (2009) Correlation of plasma insulin concentration with laminitis score in a field study of equine cushing's disease and equine metabolic syndrome. Journal of Equine Veterinary Science 29(2): 87-94.

41 12. McGowan CM, Frost R, Pfeiffer DU, Neiger R. (2004) Serum insulin concentrations in horses with equine cushing's syndrome: Response to a cortisol inhibitor and prognostic value. Equine Vet J 36(3): 295-298.

13. Frank N, Elliott SB, Chameroy KA, Toth F, Chumbler NS, et al. (2010) Association of season and pasture grazing with blood hormone and metabolite concentrations in horses with presumed pituitary pars intermedia dysfunction. J Vet Intern Med 24(5): 1167-1175.

14. Treiber KH, Kronfeld DS, Hess TM, Byrd BM, Splan RK, et al. (2006) Evaluation of genetic and metabolic predispositions and nutritional risk factors for pasture-associated laminitis in ponies. J Am Vet Med Assoc 228(10): 1538-1545.

15. Frank N. (2009) Equine metabolic syndrome. Journal of Equine Veterinary Science 29(5): 259-267.

16. Brooks SA, Gabreski N, Miller D, Brisbin A, Brown HE, et al. (2010) Whole-genome SNP association in the horse: Identification of a deletion in myosin va responsible for lavender foal syndrome. PLoS Genet 6(4): e1000909.

17. Lykkjen S, Dolvik NI, McCue ME, Rendahl AK, Mickelson JR, et al. (2010) Genome- wide association analysis of osteochondrosis of the tibiotarsal joint in norwegian standardbred trotters. Anim Genet 41: 111-120.

18. Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, et al. (2007) PLINK: A tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet 81(3): 559-575.

19. Kent WJ, Sugnet CW, Furey TS, Roskin KM, Pringle TH, et al. (2002) The human genome browser at UCSC. Genome Res 12(6): 996-1006.

20. Peters T, Ausmeier K, Ruther U. (1999) Cloning of fatso (fto), a novel gene deleted by the fused toes (ft) mouse mutation. Mamm Genome 10(10): 983-986.

21. Frayling TM, Timpson NJ, Weedon MN, Zeggini E, Freathy RM, et al. (2007) A common variant in the FTO gene is associated with body mass index and predisposes to childhood and adult obesity. Science 316(5826): 889-894.

22. Peeters A, Beckers S, Verrijken A, Roevens P, Peeters P, et al. (2008) Variants in the FTO gene are associated with common obesity in the belgian population. Mol Genet Metab 93(4): 481-484.

23. Cha SW, Choi SM, Kim KS, Park BL, Kim JR, et al. (2008) Replication of genetic effects of FTO polymorphisms on BMI in a korean population. Obesity (Silver Spring) 16(9): 2187-2189.

42 24. Adeyemo A, Chen G, Zhou J, Shriner D, Doumatey A, et al. (2010) FTO genetic variation and association with obesity in west africans and african americans. Diabetes 59(6): 1549- 1554.

25. Bravard A, Lefai E, Meugnier E, Pesenti S, Disse E, et al. (2011) FTO is increased in muscle during type 2 diabetes, and its overexpression in myotubes alters insulin signaling, enhances lipogenesis and ROS production, and induces mitochondrial dysfunction. Diabetes 60(1): 258-268.

26. Lappalainen T, Kolehmainen M, Schwab U, Pulkkinen L, de Mello VD, et al. (2010) Gene expression of FTO in human subcutaneous adipose tissue, peripheral blood mononuclear cells and adipocyte cell line. J Nutrigenet Nutrigenomics 3(1): 37-45.

27. Church C, Moir L, McMurray F, Girard C, Banks GT, et al. (2010) Overexpression of fto leads to increased food intake and results in obesity. Nat Genet 42(12): 1086-1092.

28. Fredriksson R, Hagglund M, Olszewski PK, Stephansson O, Jacobsson JA, et al. (2008) The obesity gene, FTO, is of ancient origin, up-regulated during food deprivation and expressed in neurons of feeding-related nuclei of the brain. Endocrinology 149(5): 2062- 2071.

29. Sladek R, Rocheleau G, Rung J, Dina C, Shen L, et al. (2007) A genome-wide association study identifies novel risk loci for type 2 diabetes. Nature 445(7130): 881-885.

30. Gunn TM, Miller KA, He L, Hyman RW, Davis RW, et al. (1999) The mouse mahogany locus encodes a transmembrane form of human attractin. Nature 398(6723): 152-156.

31. He L, Gunn TM, Bouley DM, Lu XY, Watson SJ, et al. (2001) A biochemical function for attractin in agouti-induced pigmentation and obesity. Nat Genet 27(1): 40-47.

32. Haqq AM, Rene P, Kishi T, Khong K, Lee CE, et al. (2003) Characterization of a novel binding partner of the melanocortin-4 receptor: Attractin-like protein. Biochem J 376(Pt 3): 595-605.

33. Muto Y, Sato K. (2003) Pivotal role of attractin in cell survival under oxidative stress in the zitter rat brain with genetic spongiform encephalopathy. Brain Res Mol Brain Res 111(1-2): 111-122.

34. Walker WP, Aradhya S, Hu C, Shen S, Zhang W, et al. (2007) Genetic analysis of attractin homologs. Genesis 45(12): 744-756.

35. Schioth HB, Muceniece R, Wikberg JE. (1996) Characterisation of the melanocortin 4 receptor by radioligand binding. Pharmacol Toxicol 79(3): 161-165.

43 36. Rossi M, Kim MS, Morgan DG, Small CJ, Edwards CM, et al. (1998) A C-terminal fragment of agouti-related protein increases feeding and antagonizes the effect of alpha- melanocyte stimulating hormone in vivo. Endocrinology 139(10): 4428-4431.

37. Kask A, Pahkla R, Irs A, Rago L, Wikberg JE, et al. (1999) Long-term administration of MC4 receptor antagonist HS014 causes hyperphagia and obesity in rats. Neuroreport 10(4): 707-711.

38. Huszar D, Lynch CA, Fairchild-Huntress V, Dunmore JH, Fang Q, et al. (1997) Targeted disruption of the melanocortin-4 receptor results in obesity in mice. Cell 88(1): 131-141.

39. Haskell-Luevano C, Schaub JW, Andreasen A, Haskell KR, Moore MC, et al. (2009) Voluntary exercise prevents the obese and diabetic metabolic syndrome of the melanocortin-4 receptor knockout mouse. FASEB J 23(2): 642-655.

40. Suagee JK, Corl BA, Hulver MW, McCutcheon LJ, Geor RJ. (2011) Effects of hyperinsulinemia on glucose and lipid transporter expression in insulin-sensitive horses. Domest Anim Endocrinol 40(3): 173-181.

41. Vervuert I, Voigt K, Hollands T, Cuddeford D, Coenen M. (2009) Effect of feeding increasing quantities of starch on glycaemic and insulinaemic responses in healthy horses. Vet J 182(1): 67-72.

42. Stewart-Hunt L, Pratt-Phillips S, McCutcheon LJ, Geor RJ. (2010) Dietary energy source and physical conditioning affect insulin sensitivity and skeletal muscle glucose metabolism in horses. Equine Vet J 42 Suppl 38: 355-360.

43. Frank N, Geor RJ, Bailey SR, Durham AE, Johnson PJ, et al. (2010) Equine metabolic syndrome. J Vet Intern Med 24(3): 467-475.

44. Stephenson HM, Green MJ, Freeman SL. (2011) Prevalence of obesity in a population of horses in the UK. Vet Rec .

45. Wyse CA, McNie KA, Tannahill VJ, Murray JK, Love S. (2008) Prevalence of obesity in riding horses in scotland. Vet Rec 162(18): 590-591.

46. Lehnhard RA, McKeever KH, Kearns CF, Beekley MD. (2004) Myosin heavy chain profiles and body composition are different in old versus young standardbred mares. Vet J 167(1): 59-66.

47. McFarlane D, Miller LM, Craig LE, Dybdal NO, Habecker PL, et al. (2005) Agreement in histologic assessments of the pituitary pars intermedia in aged horses. Am J Vet Res 66(12): 2055-2059.

48. van der Kolk JH, Heinrichs M, van Amerongen JD, Stooker RC, in de Wal LJ, et al. (2004) Evaluation of pituitary gland anatomy and histopathologic findings in clinically

44 normal horses and horses and ponies with pituitary pars intermedia adenoma. Am J Vet Res 65(12): 1701-1707.

49. Dauer W, Przedborski S. (2003) Parkinson's disease: Mechanisms and models. Neuron 39(6): 889-909.

50. Arnold G, Gasser T, Storch A, Lipp A, Kupsch A, et al. (2005) High doses of pergolide improve clinical global impression in advanced parkinson's disease:- a preliminary open label study. Arch Gerontol Geriatr 41(3): 239-253.

45 APPENDIX For each phenotypic classification association statistics were calculated. Notable associations were discussed in the main body of this work. For comparison, results from all calculations are shown below. All Manhattan plots have a reference line to show the significance threshold for multiple testing.

Supplemental Figure 1: Manhattan Plot of Qualitative Association of BMP (48 horses) vs. N (17 horses). Horses with any diagnosis compared to normal, healthy horses.

Supplemental Figure 2 Quantile-quantile Plot of Qualitative Association (48 horses) vs. N (17 horses) 46

Supplemental Figure 3 Manhattan Plot of Qualitative Association of BP (33 horses) vs. MN (32 horses). Horses with elevated ACTH levels compared to those without elevated ACTH levels.

Supplemental Figure 4 Quantile-quantile Plot of Qualitative Association BP (33 horses) vs. MN (32 horses).

Supplemental Figure 5 Manhattan Plot of Qualitative Association of BM (24 horses) vs. PN (41 horses). Horses with elevated insulin levels compared to those without elevated insulin levels.

Supplemental Figure 6 Quantile-quantile Plot of Qualitative Association BM (24 horses) vs. PN (41 horses).

Supplemental Figure 7 Manhattan Plot of Qualitative Association of B (9 horses) vs. PMN (56 horses). Horses with elevated levels of ACTH and insulin compared to normal, healthy controls and those with either ACTH or insulin levels elevated.

Supplemental Figure 8 Quantile-quantile Plot of Qualitative Association B (9 horses) vs. PMN (56 horses).

Supplemental Figure 9 Manhattan Plot of Qualitative Association of M (15 horses) vs. N (17 horses). Horses with elevated insulin levels only compared to normal, healthy controls.

Supplemental Figure 10 Quantile-quantile Plot of Qualitative Association M (15 horses) vs. N (17 horses).

Supplemental Figure 11 Manhattan Plot of Qualitative Association of BM (24 horses) vs. N (17 horses). Horses with elevated insulin levels compared to normal, healthy controls.

Supplemental Figure 12 Quantile-quantile Plot of Qualitative Association BM (24 horses) vs. N (17 horses).

Supplemental Figure 13 Manhattan Plot of Qualitative Association of M (15 horses vs. BPN (50 horses). Horses with elevated insulin levels only compared to healthy horses and those with elevated ACTH levels.

Supplemental Figure 14 Quantile-quantile Plot of Qualitative Association M (15 horses vs. BPN (50 horses).

Supplemental Figure 15 Manhattan Plot of Qualitative Association of BP (33 horses) vs. N (17 horses). Horses with elevated ACTH levels compared to normal, healthy controls.

Supplemental Figure 16 Quantile-quantile Plot of Qualitative Association BP (33 horses) vs. N (17 horses).

Supplemental Figure 17 Manhattan Plot of Qualitative Association of B (9 horses) vs. N (17 horses). Horses with elevated levels of ACTH and insulin compared to normal, healthy controls.

Supplemental Figure 18 Quantile-quantile Plot of Qualitative Association B (9 horses) vs. N (17 horses).

Supplemental Figure 19 Manhattan Plot of Qualitative Association of P (24 horses) vs. N (17 horses). Horses with elevated ACTH levels only compared to normal, healthy controls.

Supplemental Figure 20 Quantile-quantile Plot of Qualitative Association P (24 horses) vs. N (17 horses).

Supplemental Figure 21 Manhattan Plot of Qualitative Association of P (24 horses) vs. BMN (41 horses). Horses with elevated ACTH levels only compared to normal, healthy controls and those with elevated insulin levels.

Supplemental Figure 22 Quantile-quantile Plot of Qualitative Association P (24 horses) vs. BMN (41 horses).

Supplemental Figure 23 Manhattan Plot of Qualitative Association of PM (39 horses) vs. N (17 horses). Horses with either elevated ACTH or insulin levels only compared to normal, healthy controls.

Supplemental Figure 24 Quantile-quantile Plot of Qualitative Association PM (39 horses) vs. N (17 horses).

Supplemental Figure 25 Manhattan Plot of Qualitative Association of BP (33 horses) vs. M (15 horses). Horses with elevated ACTH levels compared to those with elevated insulin levels only.

Supplemental Figure 26 Quantile-quantile Plot of Qualitative Association BP (33 horses) vs. M (15 horses).

Supplemental Figure 27 Manhattan Plot of Qualitative Association of BM (24 horses) vs. P (24 horses). Horses with elevated insulin levels compared to those with elevated ACTH levels only.

Supplemental Figure 28 Quantile-quantile Plot of Qualitative Association BM (24 horses) vs. P (24 horses).

Supplemental Figure 29 Manhattan Plot of Quantitative Association with Highest Recorded Insulin.

Supplemental Figure 30 Quantile-quantile Plot of Quantitative Association with Highest Recorded Insulin.

Supplemental Figure 31 Manhattan Plot of Quantitative Association with Highest Recorded ACTH.

Supplemental Figure 32 Quantile-quantile Plot of Quantitative Association with Highest Recorded ACTH.

Supplemental Figure 33 Manhattan Plot of Qualitative Association with History of Laminitis (26 cases, 39 controls).

Supplemental Figure 34 Quantile-quantile Plot of Quantitative Association with History of Laminitis (26 cases, 39 controls).

62 Supplemental Table 1 Table of Significant SNPs from All Associations Association N cases/ # Loci/ Total sig Most Best Nearby controls Locus SNPs significant P-value Genes info BPMvN 48/17 3 4 ECA5 2 BIEC2- 1.83 e-5 HFM1 73.1Mb 918417 ECA10 1 BIEC2-95500 1.13 e-5 PEPD 4.6Mb ECA21 1 BIEC2- 2.53 e-5 Eps1511 1.7Mb 547188 CALR3 CHERP BPvMN 33/32 0 0 BMvPN 24/41 1 1 ECA5 1 BIEC2- 2.1 e-5 No gene 76.2Mb 919693 BvPMN 9/56 10 10 ECA5 1 BIEC2- 7.11 e-6 HMCN1 22.6Mb 898132 AGPAT2 ECA5 1 BIEC2- 1.09 e-5 No gene 22.9Mb 898144 ECA5 1 BIEC2- 1.1 e-5 CD1D 39.7Mb 906312 ECA5 1 BIEC2- 1.34 e-5 CXorf40B 65.0Mb 914202 ECA13 1 BIEC2- 1.75 e-5 CLEC16A 33.3Mb 232141 CG12753 ECA14 1 BIEC2- 1.08 e-5 AP3B1 87.3Mb 273090 ECA14 1 BIEC2- 1.47 e-5 No gene 92.1Mb 276910 ECA15 1 BIEC2- 1.26 e-5 No gene 78.7Mb 321447 ECA16 1 BIEC2- 1.24 e-5 DAZL 65.0Mb 356433 ECAX 1 BIEC2- 1.11 e-5 HUWE1 44.8Mb 1123081 MIRLET7F2 MIR98 MvN 15v17 1 1 ECA27 1 BIEC2- 1.83 e-5 No gene 28.1Mb 713528 BMvN 24/17 1 1 ECA22 1 BIEC2- 1.6 e-5 BCAS4 38.8Mb 596175 PARD6B MvBPN 15/50 0 0 N/A N/A N/A BPvN 33/17 2 2 63 ECA9 1 BIEC2- 2.26 e-5 SFT2D1 58.8Mb 1097793 ECA10 1 BIEC2-95500 3.63 e-6 PEPD 4.6Mb BvN 9/17 4 4 ECA1 1 BIEC2-42874 2.38 e-5 No gene 99.6Mb ECA1 1 BIEC2-42897 2.38 e-5 No gene 99.6Mb ECA17 1 BIEC2- 1.82 e-5 GPC6 63.4Mb 382291 ECA23 1 BIEC2- 7.01 e-6 No gene 48.8Mb 626348 PvN 24/17 3 3 ECA4 1 BIEC2- 2.58 e-5 SKAP2 57.8Mb 867420 ECA9 1 BIEC2- 5.11 e-6 SFT2D1 58.8Mb 1097793 ECA10 1 BIEC2-95500 7.07 e-6 PEPD 4.6Mb PvBMN 24/41 4 4 ECA3 1 BIEC2- 2.03 e-5 No gene 68.9Mb 789598 ECA4 1 BIEC2- 2.36 e-5 SKAP2 57.8Mb 867420 ECA5 1 BIEC2- 1.69 e-5 CACNA1E 18.9Mb 897199 ECA5 1 BIEC2- 1.01 e-5 CD1D 39.7Mb 906312 PMvN 39/17 3 3 ECA9 1 BIEC2- 1.37 e-5 SFT2D1 58.8Mb 1097793 ECA10 1 BIEC2-95500 2.32 e-5 PEPD 4.6Mb ECA16 1 BIEC2- 2.23 e-5 LZTFL1 40.8Mb 342954 SLC6A20 CCR9 BPvM 33/15 0 0 N/A N/A N/A BMvP 24/24 2 2 ECA5 1 BIEC2- 1.47 e-5 CACNA1E 18.9Mb 897199 ECA5 1 BIEC2- 1.49 e-5 CD1D 39.7Mb 906312 Highest N/A 30 38 Insulin

64 ECA 1 2 BIEC2-78264 6.10 e-6 No gene 164.2Mb ECA3 1 BIEC2- 6.10 e-6 TK2 17.1Mb 773422 CKLF CMTM1 CMTM2 BEAN1 ECA3 2 BIEC2- 6.10 e-6 LDB1 106.8Mb 808839 LDB2 ECA5 1 BIEC2- 6.10 e-6 SELP 6.1Mb 890054 SELL ECA9 2 BIEC2- 6.92 e-6 TMEM74 53.2Mb 1094996 ECA10 1 BIEC2- 5.18 e-6 PGLYRP1 16.4Mb 104865 CCDC61 NOVA2 EBAG9 ECA10 1 BIEC2- 6.10 e-6 DCBLD1 65.9Mb 130168 GOPC ECA11 1 BIEC2- 7.28 e-6 VPS53 44.7Mb 155077 ECA11 1 BIEC2- 6.10 e-6 No gene 48.9Mb 157582 ECA12 1 BIEC2- 1.48 e-6 ALX4 9.3Mb 173131 PRRX2 ECA13 1 BIEC2- 6.10 e-6 No gene 37.7Mb 235337 ECA13 1 BIEC2- 4.67 e-6 FLYWCH1 39.8Mb 235746 KREMEN2 PAQR4 PKMYT1 CLDN6 ECA14 1 BIEC2- 2.18 e-5 No gene 69.3Mb 263524 ECA14 1 BIEC2- 2.15 e-5 No gene 75.6Mb 266578 ECA15 1 BIEC2- 3.58 e-6 ANAPC1 14.9Mb 289514 ECA15 1 BIEC2- 8.41 e-6 MERTK 15.0Mb 289674 ECA15 1 BIEC2- 1.20 e-5 TGFA 32.2Mb 301455 ECA15 1 BIEC2- 6.10 e-6 BCL11B 40.6Mb 304398 ECA16 2 BIEC2- 6.10 e-6 TMEM40 4.6Mb 328336 RAF1 65 ECA17 1 BIEC2- 6.10 e-6 ABCC4 65.2Mb 382723 ECA18 2 BIEC2- 4.33 e-6 No gene 35.0Mb 410728 ECA18 1 BIEC2- 6.10 e-6 No gene 48.1Mb 412288 ECA18 2 BIEC2- 6.10 e-6 FUCA1 55.2Mb 414366 ECA18 2 BIEC2- 6.10 e-6 No gene 78.7Mb 421249 ECA18 1 BIEC2- 7.11 e-6 No gene 81.4Mb 421894 ECA22 1 BIEC2- 6.10 e-6 No gene 16.4Mb 583538 ECA23 1 BIEC2- 6.10 e-6 PPIB 28.6Mb 622442 ECA25 1 BIEC2- 7.15 e-6 ZFP933 8.2Mb 657342 ECA26 2 BIEC2- 6.10 e-6 ROBO2 12.3Mb 684068 ECA29 1 BIEC2- 2.57 e-5 SFMBT2 26.8Mb 761473 Highest N/A 3 3 ACTH ECA7 1 BIEC2- 2.40 e-5 OLFML1 76.9Mb 1007280 SYT9 PPFIBP2 ECA13 1 BIEC2- 4.34 e-7 XPO6 20.2Mb 215377 ECA17 1 BIEC2- 3.55 e-6 No gene 50.8Mb 377685 Laminitis 26/39 2 2 ECA14 1 BIEC2- 1.73 e-5 FAM174A 69.1Mb 263373 ECA14 1 BIEC2- 2.56 e-6 No gene 69.3Mb 263524